Mechanisms causing wear in a homogenizer gap

(1)

2010:018 CIV

M A S T E R ' S T H E S I S

Mechanisms Causing Wear in a Homogenizer Gap

Fredrik Forsberg Erik Hultman

Luleå University of Technology MSc Programmes in Engineering

Mechanical Engineering

Department of Applied Physics and Mechanical Engineering Division of Machine Elements

(2)

Abstract

Homogenization is the process used to disrupt fat globules in diary products and beverages. This is done to reduce the creation of cream layer (separation) and to

enhance the viscosity of certain products. This process takes place in a narrow gap in the homogenizer machine. The process causes wear in the homogenization gap and after an amount of wear time the machine loses its homogenization effect.

This thesis aim is to identify the mechanisms causing wear in the homogenization gap and to find a measurement method so that experiments can be carried out. The wear is analyzed by measurement of the weight loss, i.e. the worn off material. To measure the weight loss a balance and an optical 3D-profiler is used. The structure of the worn surface is also analyzed using the microscope pictures from the optical 3D-profiler.

Two experimental rigs were used to study the differences between wear during operating conditions with and without particles. For the tests without added particles the influence of cavitation-erosion phenomena was studied. For the tests with added particles the influence of particle properties and the effect of cavitating and

non-caviting conditions were studied. The trajectories of the particles are simulated without the influence of cavitation using a CFD-code, to see if the particles could be a possible cause of wear. The gap has been simulated for a worn geometry in order to see how the particle trajectories change.

The results show that added particles accelerate the wear compared to that in tests without added particles. The most important parameters affecting the wear are the particle hardness and mass. When cavitation and particles are combined they create a synergetic effect on the wear amount. This can be explained by the fact that cavitation can accelerate particles in random directions by the action of the implosion of the cavity. According to the theory of erosion wear a changed impact angle, from low to medium and increased velocity will increase the wear amount significantly.

CFD-simulations and calculations show that the particles do not follow the streamlines well and therefore creates wear on the gap surfaces. The simulations of the particle trajectories show that the changed geometry due to wear changes the velocities and impact angles of the particles thus creating more wear.

(3)

1 Acknowledgements

This diploma work is carried out as the last step in the educational program in

mechanical engineering at Luleå University of technology. The work is carried out during 20 weeks corresponding to 30 university points.

We would like to thank our supervisors at Tetra Pak Processing Systems:

• Fredrik Innings

• Rolf Malmberg

• Pavlos Kouroutsidis

And our examiner and supervisor at the university:

• Braham Prakash

Also many thanks to other persons that have contributed to our work:

• Fredrik Johansson, Erik Börjesson, Lars Hamberg, Thomas Skoglund, Göran Pantzar and all the other employees at Tetra pak processing systems

• Kurt Glock and the employees in the workshop for their help and input to our work.

• The employees at PDC for the help with the experimental rig.

• Anita Lindbäck and Philippe Lingois at materials technology for borrowing their balance and optical surface profiler.

(4)

2 Background

This project was initiated in order to get a fundamental understanding of the

mechanisms causing wear in the gap geometry of the homogenizer machine Tetra Alex.

The mechanisms have not been investigated before therefore it is crucial to get a fundamental understanding of the phenomena’s causing wear. The main questions are:

• How to measure wear?

• Which mechanisms cause the wear?

• Where is the wear located in the gap?

• How do the flow field and the geometry influence the wear?

• What could be done to reduce the wear?

The products known to create most wear in production are for instance; milk, chocolate flavored milk, calcium-enriched milk, soya milk and ketchup. Approximate values of the production time for milk is 20 000 h, for calcium enriched milk and juice 1000 h and for ketchup 0,5 h, the worn surface for ketchup is shown in Figure 1.

Figure 1: The wear pattern created on the forcer by tomato paste after 30 minutes homogenization. The material is Wallex 20.

(7)

3 Objectives and scope of work 3.1 Objectives

Four main objectives have been identified in this diploma work as listed below.

Define methods to measure wear

• Investigate the position of the wear

• Investigate different wear types and their characteristics

• Find a measure to quantify the amount of wear What mechanisms causes wear?

How does cavitation affect wear?

• Investigate how the cavitation wear zone position is affected by the Thoma number

• Identify characteristics of the wear pattern created by cavitation

• Measure the weight loss depending on Thoma number

What is the effect of particles upon wear?

• Investigate if a certain wear pattern is created by particles

• Investigate the difference on wear depending on particle properties

• Investigate the combined effect of cavitation and particles

What is the relation between machine parameters and wear?

• Investigate how the position of the wear zone changes with the gap height

• Investigate the wear pattern depending on different machine parameters

• Investigate the weight loss depending on machine parameters How does the geometry and flow field affect the wear?

• Investigate if there is a connection between the flow field and the wear pattern

• Investigate how the flow field is affected by a geometrical change caused by wear

(8)

3.2 Scope of work

The wear is studied for the gap region in homogenizers.

The following operational conditions have been studied:

• Calcium particles and low cavitation with varying gap heights, 5 runs

• Corundum particles and low cavitation, 3 and 10μm, 2 runs

• Calcium particles and high cavitation, 1 run

• Calcium particles without cavitation, 1 run

• Municipal water with varying Thoma number for three machines (200, 400 and 600bar), 8 runs

The weight loss is measured for all samples. The duration of experiments with varying gap height and calcium particles is 2 hours.

For the gap height with corundum the weight loss is measured after 3 min and 30 min.

For cavitation and particles and particle wear the time evolution of wear is studied during 2-3 hours. Tests with varying Thoma number is run for 50-550 hours depending on operating conditions and machine type. The weight loss is only measured before and after the total wear time for the sample, i.e. the evolution of the wear with time is not studied.

All samples in the list above except cavitation with municipal water and two of the gap heights samples are measured with an optical profiler. One position at each sample is measured and is chosen to represent the characteristic wear pattern for that sample.

CFD-simulations are carried out for the worn surfaces that have been scanned with the optical 3-d profiler. The unworn surface profile and the profile for calcium particles and high cavitation are simulated.

Wear from mainly water suspensions are studied, the particles used in the experiments are corundum and calcium particles.

Wear for one viscous fluid, tomato paste, is studied.

(9)

4 Homogenizer Tetra Alex

4.1 Homogenization

Homogenization is the process that is used to disrupt fat globules in dairy products and beverages. This is done to reduce creation of cream layer (separation), create a more appetizing color and satisfying consistence at ingestion. A principal picture of the reduction in size is seen in Figure 2.

Figure 2 Picture A shows the size of the fat globules before homogenization and picture B shows the size after homogenization.

The homogenization is done in a homogenizer which is a machine consisting of several parts which are listed in Table 1 and can be seen in Figure 3.

Table 1: The main parts of a homogenizator

Figure number Part name

1 Main drive motor

2 V-belt transmission

3 Gearbox

4 Damper

5 Hydraulic pressure setting system 6 Homogenizing device, second stage 7 Homogenizing device, first stage 8 Solid stainless steel pump block

9 Pistons

10 Crankcase

(10)

Figure 3: Picture of a homogenizator with its essential parts, the location of the gaps is in 6 and 7. The diploma work concerns wear in the gap located in 7. [From Dairy processing handbook]

The gap located in the homgenization device is the area where the actual

homgenization is created. The gap consists of a forcer shaft and a seat, seen in Figure 4.

The forcer shafts in Figure 5 are of the old HD1 standard, the newer standard is HD100.

See Figure 4. There are several modifications made between the two standards the hydraulic system, the homogenization device and the geometry of the gap.

Figure 4: HD100-gap the disc on the shaft is turn able.

The forcer shaft exists in two configurations, one is solid and one has a turn able disk.

The turn able disk results in a longer lifetime due to the fact that both sides of the disk can be used before replacement.

Forcer shaft Forcer

Seat

(11)

4.2 Operation of the homogenizer

The product to be homogenized flows through the narrow gap, which is created by exerting a force on the forcer shaft. The magnitude of the force acting on the forcer is controlled by change of the hydraulic pressure; see Table 2^X and Figure 5.

The force balance involves three forces:

1. The pressure force exerted by the fluid on the front of the forcer 2. The pressure on the back of the forcer from the second stage pressure 3. The hydraulic pressure exerted on the forcer

The gap height is adjusted by changing the first stage hydraulic pressure or the flow rate.

Table 2: Components for a two-stage homogenization machine

No Part

1 First stage forcer 2 Second stage forcer 3 Seat

4 Gap

5 Hydraulic actuator

Figure 5: The components of the homogenization devices, the area in the gap region are encircled in the figure. [From Dairy processing handbook]

(12)

Fluid is forced through this gap with high velocity to homogenize it. According to Bernoulli’s equation a high velocity is accompanied by a low static pressure. In the gap the static pressure is below the vapor pressure, this causes the formation of cavitation, see ^X5.3 Cavitation.

The intensity of the cavitation depends on the Thoma number, Th, of the second stage pressure (P2) to the first stage pressure (P1), Th=P2/P1. Cavitation creates sound and light and thus it is possible to measure. The experiments show that for Th-values from 0 to 0,05 the light and sound intensity emitted by the cavitation bubbles is much higher than between 0,05 and 0,15 where the intensities drop very quickly^F¹.

The pressure and velocity variations are shown principally in ^XFigure 6^X for two different cases. The solid line represents Th=0 and corresponds to the case when the second stage pressure is not pressurized, Th=0,6 corresponds to the case when second stage pressure is 60 % of the first stage pressure. Firstly the pressure drops from P1 to a low level in the gap as a consequence of the increase in velocity. Since the flow rate is constant in formula 1, the velocity increases when the area reduces. In the expansion the pressure recovers to the pressure level at the second stage, P2, when the velocity reduces. Since the pressure losses is fairly constant, the pressure level in the gap will be higher for Th=0,6 compared to Th=0 in Figure 6.

Figure 6: The gap is illustrated in 2-d. The solid and dashed lines show how the pressure and velocity varies in the gap region.

TP

1

PT Håkansson, A. (2008). Visual Observation and Acoustical Measurements of Cavitation in a High Pressure Homogenizer Model. Lund.

(13)

Th=0 – High cavitation Th=0,1 – Medium cavitation Th=0,2 – Low cavitation Th=0,6 – No cavitation

In Figure 7 an overview of the pump is shown, during one revolution of the crankcase the fluid is first sucked in during the suction stroke. When the suction stroke is in the end phase the piston chamber is filled with the liquid to be pressurized. Now the piston changes direction and the pressure stroke starts. The fluid is pressurized in the chamber and when the pressure force on the upper mushroom valve is larger than the

homogenizing pressure and the spring stiffness force the fluid will flow through the upper mushroom valve.

Figure 7: The suction stroke in the chamber to the right and the pressurization in the chamber to the left. The pistons act into and out of the paper depending on if they are on the suction or pressure

stroke. [Tetra pak, homogenization dept.]

Since the homogenizator uses a piston pump to displace the fluid the pressure will vary at the inlet as a periodic function. This of course gives a displacement of the forcer that oscillates with the frequency of the pressure variation.

Pistonchamber Mushroomvalve Pressurized fluid

(14)

5 Theory

5.1 Particles in food

Food contains particles with different properties and some of these are hard and rather large and therefore can cause potential damage to surfaces. There are many fluids being processed in the homogenizer and many of those contains particles with may cause damage to surface by the impact on a surface.

The main reason for the study of the particles suspended in the liquid is to determine the shape, hardness and density of the suspended particles. With knowledge of the size two cases of possible damage cases can be distinguished, when the particles are smaller or bigger than the gap height. In the case of particles smaller than the gap height the flow field will determine how they will impact the gap surfaces and eventually cause damage. If the particle is bigger than the gap height it will be forced through the gap and cause formation of grooves in both forcer and seat.

The gap is where the function of the machine is created; therefore it is crucial to understand how to design it for a longer lifetime. This would enable the costumers to produce with less interruption in the process due to service of the machine.

In pear for example there are particles with hard cell walls of lignin called stone cells, see Figure 8.

Figure 8: Measurements of pear puré with flow cam.

(15)

5.2 Calculations Gap height

Figure 9: The gap geometry in 2-d.

The gap height is calculated using continuity of mass:

Q=V₁A₁=V₂A₂ ⁽¹⁾

The cross sectional area at 2 is given by A₂=π^d2h ⁽²⁾

Using the geometries in Figure 9, with point 1 at the inlet and 2 in the low pressure region in the gap, i.e. P^B2^B is neglected. The Bernoulli equation is used to determine the pressure and velocity at different positions along a streamline between point 1 and 2.

P₁+ρ^V¹²

2 =P₂+ρ^V²²

2 (3)

From (3) it can be concluded that the velocity at point 2 in Figure 9 is given by:

V₂= 2 P

(

₁−P₂

)

ρ (4)

(16)

Using (4) and (2) inserted in (1) and solved for the gap height gives:

h= Q 2 P

(

₁−P₂

)

ρ

1 π^d2

(5)

Where P1 is the total homogenization pressure.

The calculations for the gap height given in (5) does not account for frictional losses. The frictional losses can be determined by the swamee-jain formula for the friction factor, given in (6).

f = 0,25 lg ε

3,7D+ 5,74 Re^0,9



 





 



2 (6)

If the gap height is calculated with pressure losses the result changes slightly, in the order of 5-10 µm.

In equation 6 ε=k/Dh is the relative roughness, where k is the absolute roughness value of the material surface 0,000015 for stainless steel. Dh=4A/C is the hydraulic diameter, A is the mantle area perpendicular to the flow direction of the fluid, C is the perimeter of the mantle area πd2. Re=ρV2L/µ is the Reynolds number. The pressure losses in the system are given by f L

D_h ρ^V2

2

, where L is the length of the distance for which the pressure drop is to be calculated. With this interpreted to equation (3) the calculation of the gap height is more accurate and the bernoullis equation accounting for frictional losses thus becomes:

P₁+ρ^V¹²

2 =P₂+ρ^V²² 2 + f L

D_h ρ^V2

2

(7)

The gap height exists in all terms in equation (7) and can be solved by inserting a gap height and then iteratively find the gap height. There exists a code for this procedure at tetra pak and this code is used to calculate the gap height.

(17)

5.3 Cavitation

The phenomenon cavitation is known to cause severe damage to hydrodynamic systems by the formation and implosion of vapor bubbles. Cavitation can be induced in many ways and most focus is on research in the field of acoustic-and hydrodynamic cavitation.

The difference between these two cases is that in the acoustic case the medium is static and the cavitation is induced by ultrasound. In the hydrodynamic case the liquid is in motion and the balance between velocity, pressure and pressure losses determines when the pressure drops below the threshold for cavitation. The threshold is the vapor pressure of the fluid. In order to analyze the dynamics of a bubble it is important to understand which forces that are involved, see Figure 10. If the internal pressure can be considered constant it is apparent that when the ambient pressure decreases the bubble will start to expand and vice versa. In most cases the bubble can be analyzed as being in equilibrium (up to the point of collapse), according to equation 8:

P_A + 2γ

R = P_V + P_g (8)

Figure 10: The pressure forces acting on a gas bubble. [From national committee for fluid mechanics]

In the theoretical literature concerning cavitation much focus is on water as a pure chemical substance without impurities. Normal tap water contains impurities, called nucleation sites. The nucleation sites are crucial for at which pressure value the

threshold of cavitation inception, i.e. the initial formation of vapor voids starts. In Figure 11 picture B shows when the liquid is in vapor phase. The curve marked actual isotherm is for water with nucleation sites and the theoretical isotherm is for pure water without nucleation sites. The actual isotherm takes the isobaric path between point B and C, which is the value of the vapor pressure of the fluid. The theoretical isotherm on the

(18)

other hand can withstand large negative pressures, i.e. tensions. This theoretical behavior of water has been showed experimentally by rigorous cleaning of the water^F²^F.

Figure 11: Phase diagrams for water shows when water exists in liquid or vapor phase depending on pressure, temperature and volume. [From Brennen C.E Cavitation and bubble dynamics]

For normal tap water vapor cavities are formed when the local pressure drop below the vapor pressure of water which is 2338 Pa at 20°C.

There are several situations where cavitation typically is observed^F³^F:

• For low pressure regions where the pressure is below the vapor pressure

• Wall geometry that gives rise to large fluid acceleration or curvature effect

• Shear flows (jets and wakes)

• Vibratory motion of walls that causes an oscillating pressure field

Normally to quantify the amount of cavitation or the aggressiveness of the cavities one uses the σ-model proposed in a review article for avoiding control valve problems⁴^F: σ = P₁−P_v

P₁−P₂

(9)

2 Q, Z., & al., D. D. (1991). Liquids at large negative pressures: Water at the homogeneous nucleation limit. Science , 254, ss. 829-832.

3 Brennen, C. E. (1995). CAVITATION AND BUBBLE DYNAMICS. Oxford University Press.

4 K.W. ROTH, J. S. (Aug 2001). Avoid control valve application problems with physics- based models. HYDROCARBON PROCESSING , ss. 37-48.

(19)

Where P₁ is the pressure upstream of the valve, P_v the vapor pressure at the actual temperature and P₂ is the downstream pressure. The smaller value the σ-value the greater cavitation potential. Cavitation inception is at a value close to 1.

The definition of the cavitation number is seldom used for the homogenizer. More common is to use the Thoma number (Th) defined as the ratio, P2/P1, as mentioned earlier. Investigations have shown that for Th-values between 0 and 0,2 cavitation is present^TF⁵^F. Best homogenization quality is obtained for the pressure ratio p2/p1=0,2^F⁶^F. 5.3.1 Collapse

Some assumptions need to be introduced to be able to derive the critical values of the radius and pressure for a cavitation bubble. Pressure change so that mechanical equilibrium, i.e. the force balance in equation 8 is still physically valid. Pressure change must be quick to ensure that diffusion of gas from bubble to dissolved gas present in surrounding is negligible. Heat transfer is possible to the surrounding. In other words this means that during the expansion of a bubble the mass of gas in the bubble is constant and isothermal. This is valid since the process of collapse is of the order 1 ms - 1 µs, which is small compared to the characteristic time for diffusion, which is 1-2 s.

Diffusion is of interest only for cavitation nuclei in equilibrium for example bubbles in a static liquid.

Cavitation bubbles can be stable or unstable. It is in the unstable case that the bubble collapse, in the stable case the bubble just expands slightly and then returns to the initial size as the ambient pressure increases. The threshold for this instability can be calculated if the initial radius is known.

Assumption, the isothermal gas law, Boyle’s law and shape of a spherical bubble:

PV = K (10)

V R( )⁼ ⁴^π^R³

3

(11)

Gives for the growth of a bubble between initial state, index 0, to arbitrary state:

P_{g 0}V₀ =P_gV ⇒ P_g = P_{g 0}V₀

( )

R₀

V R

( )

⁽¹²⁾

Finally, with (12) in (8):

P_A = P_g R₀ R



 



3

+ P_v − 2γ R

(13)

5 Håkansson, A. (2008). Visual Observation and Acoustical Measurements of Cavitation in a High Pressure Homogenizer Model. Lund.

6 Diary processing handbook. Lund: Tetra Pak Processing Systems.

(20)

The minimum value of (13) is of interest since this is the critical value mentioned above.

dP_A dR _R₌_R

C

=0 in equation 13 gives the critical values.

R_C =R₀ 3P_{g 0}R₀

2γ and _P

c = P_v− 4γ 3R_c

(14 a, 14 b)

Equations 14a and 14b gives the values for where the bubbles gets instable and will collapse. For values of the pressure below Pc the bubble will grow quickly and get much larger than it was initially and a bubble collapse will follow. The bubble collapse will cause the emission of a shock wave and a jet. For values larger than Pc the bubble grows and then returns to the initial state when the pressure recovers. If the initial radius of the vapor bubble is known the critical radius and critical pressure can be determined, giving the limit for cavitation implosion.

5.3.2 Cavitation erosion

Cavitation erosion is the damage caused by the implosion of the vapor bubbles on a metal surface causing localized stresses of high amplitude. The surface pressures created by the implosion of a vapor bubble is not static but periodic, this affects the yield strength and breaking limit of the material negatively. In the collapse of a cavitation bubble several phases has been showed to be present. Which one of these mechanisms causing the damage has been debated over the years and initially the focus was on the shock wave created by the bubble collapse. Later it was found that a

microjet is created after the shock wave and much of the focus was on to study this.

Afterwards followed research in order to find out if it is the microjet or the shockwave that creates the large pressures on the surface of the material causing the surface fatigue. Many experimentalists conducted experiments where the pressure was monitored and triggered to a camera in order to see which phenomena that were causing the largest pressures. Two independent experiments showed that the shock wave was primary responsible for the high-pressure levels.

The peak pressure in the fluid can be estimated by using the relation:

P_p=100R_mP_∞

r (15)

R^Bm is the maximum radius of the bubble; P∞ is the far field pressure, which is the actual pressure in the fluid surrounding the bubble; r is the distance from the wall. As can be concluded from equation 15 the pressure is highest near the walls and then decays rapidly to lower levels.

The collapse of a cavitation bubble is a transient event and produces very high pressures localized to a very small area. In Figure 12 the schematic growth and collapse of a vapor bubble is seen. In Figure 13 the wear pattern of cavitation-erosion is seen.

(21)

Figure 13: Typical wear pattern for cavitation erosion, the surface is very rough and has a porous structure at the surface. [From ASM Handbook]

Figure 12: The implosion of a vapor bubble causing the cavitation erosion by emission of a shockwave.

(22)

5.4 Erosive wear

In Figure 14 the parameters governing erosion wear is seen. And from Figure 15 the amount of erosion can be determined by knowledge of the incidence angle and if the surface material is ductile or brittle.

Figure 14: Erosion wear is defined as particles impinging on a surface with an incident angle and velocity. [from K-H Zum Gahr, microstructure and wear of materials]

Figure 15: The erosion amount depending on impact angle for a soft and ductile and a hard brittle material respectively. [from K-H Zum Gahr, microstructure and wear of materials]

(23)

5.5 Stokes number

The stokes number, see equation 16, is a measure of how closely a particle is following the streamlines in the flow field. For Stk>0,25 particles impact surface and for Stk<0,25 the particles don’t impact the surface. The stokes number is defined as:

Stk=ρpD_p²V

18µ^L ⁽¹⁶⁾

The characteristic length, L in equation 16, is chosen to be the hydraulic diameter as defined in equation 17.

D_h = 4 A

P = 4π^d2h

2πd₂ = 2 h (17)

The stokes numbers for calcium and corundum with ρ^Bcalcium^B=1550 kg/m³, ρ^BCorundum^B=4000 kg/m³,Dp=10 µm, µ=1040 µPa-s, L=DH=2h=100 µm V=120 m/s:

Stk^Bcalcium^B=10 Stk^Bcorundum^B=26

5.6 Estimation of particle wear

Surface pressures due to particle impacts have been shown to successfully correlate with the water hammer formula, equation 18.

P_wh = ρpc_pU_I 1+ ρpc_p

ρmc_m

(18)

ρp=Density of particle ρm=Density of surface material

cp=Speed of sound in the particle material cm=Speed of sound in the surface material U_I=Impact speed normal to the surface

For fluid and jet impingement ρ is the fluid density, c the speed of sound in the fluid and UN the particle velocity normal to the surface. The impact pressure is given by the water hammer formula, in equation 18. It was concluded that if the surface pressure according to the water hammer pressure, was in the range of 1/3 to 1/2 of the material fatigue

(24)

strength damage occurred⁷. Therefore it is possible to calculate the maximum normal velocity by inserting 1/3 to 1/2 of the material fatigue strength, P_wh.

Calcium Corundum, Al₂O₃

ρp=1550 Kg/m³ ρp=4000 Kg/m³

c_p=3810 m/s c_p=5000 m/s

Stainless steel, SS 2333 Stellite 1

Pwh=83,3 Mpa Pwh=220 Mpa

ρm=7800 Kg/m³ ρm=8690 Kg/m³

cm=5090 m/s cm=5090 m/s

P_wh,Ca =16 m/s P_wh,Ca =42 m/s

P_{wh ,Al}

2O₃ =6, 3 m/s P_{wh ,Al}

2O₃ =16 m/s

This indicates the threshold velocities for when damage will occur but does not give an time estimation for the time until fracture.

5.7 Particle acceleration caused by cavitation

Accelerated surface erosion has been observed for flow situations where particles and cavitation are present at the same time. The effect on the wear was way beyond the erosive effect by that for just particles or cavitation alone. The cavitation phenomena’s impact on particles was investigated experimentally and numerically⁸^F. The result found out was that “particles which serve as nucleation sites for cavitation-bubbles, are set into fast translating motion during the explosive growth of the cavity”. This can be seen in Figure 16 by looking at the evolution of the acceleration. The particle in the initial state can be seen in frame 1 (0.0 µs). Afterwards a cavity start to grow at the particle (1.0 µs-9.8 µs) until the particle is de-attached from the cavity (9.8 µs-14,7µs) and is being accelerated. During the experiment high velocities for the particles was observed.

TP

7

PT THIRUVENGADAM A, S. R. (June 1969). Experimental and Analytical

Investigations on Liquid Impact Erosion. ASTM Special Technical publication , 474, ss.

249-287.

TP

8

PT BRAM M.BORKENT, M. A.-D. (2008). The acceleration of solid particles subjected to cavitation nucleation. Journal of fluid mechanics , 610, ss. 157-182.

(25)

Figure 16: The solid particle acceleration observed [From Bram M.Borkent et al., “The acceleration of solid particles subjected to cavitation nucleation”]

Figure 17: The damage caused at the surface of a impeller pump under cavitating conditions with particles in the liquid. [From ASM Handbook]

(26)

5.8 Material theory

^F⁹

5.8.1 Work hardening

Work hardening is a process used for hardening alloys and metals. Work hardening occurs when the material is exposed to an external force, for instance, rolling. The hardening is due to dislocation multiplication and dislocation interactions. A dislocation is a line defect in the material, see Figure 18.

The phenomena dislocation describes the deformation behavior of metals. General metals have dislocations in the crystal structure, allowing slip by motion of few atoms at a given instance.

Figure 18: The slip motion of a dislocation in a crystal lattice, the dislocation slips from a position in (a) to the edge in (b), the b is the Burgers vector, it is defined as the unit displacement of a dislocation.

(Picture taken from ASM Handbook)

Slip in a crystal structure occurs on a specific crystal plane and in specific directions on a slip plane. The plane and direction for slip within a crystal lattice depends on how the atoms are packed. Slip planes are often on the closest-packed or closely packed plane.

These planes have the widest spacing between dislocations, and thus the dislocations move easier in these planes, see Figure 19.

9 ASM Handbook

Figure 19 Three common types of crystal structures, a) hexagonal close-packed, b) face-centered cubic, c) body-centered cubic, a) and b) are close-packed structures. (Pictures is taken from ASM Handbook)

a b c

(27)

When the material work hardens the dislocations move on intersecting slip systems collide and a number of these will become pinned. The dislocation density increases with the amount of plastic deformation applied, resulting in limited dislocation motion, which prevents dislocations gliding on intersecting systems. Consequently, the strength also increases.

5.8.2 Material properties

Stainless steel

The stainless steel used in the experiments is SS 2333 as mentioned earlier. This is an austenitic material and ductile metal.

Cobalt-Base Alloy

The cobalt-base alloy Tetra Pak use for the gap is Wallex 20. This metal has high wear, corrosion and heat resistance. The properties for alloys comes from the crystallographic nature of cobalt (in particular its response to stress).

No datasheet with tensile properties could be found for this material. To get an idea of the properties Stellite 1 is used because it was closest by hardness data and composition with an available data sheet. Wallex 20 is still the slightly harder material, but this gives a hint in how high the tensile properties for Wallex 20 are.

These two materials can be compared in Table 3 showing composition and Table 4 showing material properties.

Table 3: Material compositions for Stainless steel, Stellite and Wallex 20. [from Damstahl and Deloro]

C Co Cr Fe Ni Mn S Si P W

SS2333 ≤0,07 - 17,5-

19,5

8,00- 10,5

2,00 0,015 ^0,045

Stellite 1 2,45 Bal 31 Max

2,5

Max 3

1 - 1 - 13

Wallex 20 2,5 Bal 33 2,5 2,5 - - 1,2 - 18

Table 4: Material properties for Stainless steel and Stellite 20. [from Damstahl and Deloro]

Ultimate tensile strength, R^Bm^B,(MPa)

Yield stress, R^Bp0,2^B, (MPa)

Elastic modulus, E, (GPa)

Hardness, HRC

SS2333 500-700 190 200 20

Stellite 1 550 - 248 50-58

(28)

6 Methods & Materials

Below are the experimental method and measurement methods listed. A detailed list of the equipment is given in Appendix.

6.1 Manufacture the gaps

The gaps used in the experiments are made of SS 2333 which is a stainless steel also known as AISI 304. Usually Tetra Pak use a cobalt base alloy called Wallex 20 with very high abrasion and corrosion resistance. Because of these qualities SS 2333 is used to obtain measurable data in a short period of time.

The gaps used in the experiments are standard designs for Tetra Alex 20, 200, 350 and 30 homogenizers. Tremek in Trelleborg manufactures the gaps used in the experiments.

A list of the drawings used during manufacture of the gaps can be viewed in Table 5.

Table 5: Drawing information for the manufactured gaps

Drawing number Article number Homogenizer Seat/Forcer Diameter

4722-6883 4722-6883 01 TA 20 Seat 27 mm

6-4722 6884 00 6-4722 6884 03 TA 20 Forcer 27 mm

6-4722 7044 00 6-4722 7044 01 TA 200 Seat 64 mm

6-4722 9018 00 6-4722 9018 01 TA 200 Forcer disc 64 mm

6-4722 7099 00 6-4722 7099 03 TA 350 Seat 64 mm

6-4722 9017 00 6-4722 9017 01 TA 350 Forcer disc 64 mm

6-4722 7069 00 6-4722 7069 03 TA 30 Seat 32 mm

6-4722 7068 00 6-4722 7068 01 TA 30 Forcer 32 mm

6.2 Gap labeling

To distinguish a gap associated with a certain experiment it is marked with an engraving tool. The tool used is the Dremel Engraver and each gap is marked with a letter for a specific test, see Figure 20.

Figure 20: Engraving tool and template used for marking the gaps

Because of the reversible seat and forcer disc one side is marked with a line to keep track of the side associated to a certain test. A line on the seat, forcer and forcer disc is

(29)

created for aligning the gap with the homogenization device. This is done to assure that the gap is aligned in the same manner when assembled. The idea of the aligning line is that it makes it possible to study the influence on wear due to the anti-symmetric geometry in the homogenization device. The aligning line is aligned with a line drawn at the top of the homogenization device to ensure that it is inserted with the same

orientation each time. The engraving is done with a template delivered with the engraver tool. The method is displayed in Figure 21 below. The side marked with the recognition line seen in Figure 21^X is side 1, i.e. R1 (see tables in appendix).

Figure 21: Engraving method for the gap with a) showing the method for seat and forcer disc and b) displaying the method for the forcer

6.3 Experimental setup

Two experimental rigs are used to carry out the experiments. For the experiments that are carried out with particles an experimental rig is assembled at PDC (product

development center), Råbyholm. For the cavitation experiments with water there exists a machine testing facility at the workshop at Bryggaregatan.

The homogenizer at PDC is a TA 20. The machine models at the workshop are TA 200, TA 350 and TA 30. The drawing numbers for the machines is given in Table 5.

For a summary of machine parameters and settings used for the experiments, see 6.3.1 and 6.3.2.

X

Align line Side recognition line

Experiment recognition letter

X

Side recognition line Align line

Experiment recognition letter

a) b)

(30)

6.3.1 Experimental setup – Municipal water with added particles

To be able to run the experiments with added particles an experimental rig is set up at PDC, see Figure 22.

The experimental rig is a closed system, which allows the fluid to be circulated thus enabling the re-use of particles. The operating conditions used in the experiments are:

• 200 liters tank filled with 100 liters of water

• The concentration of particles is set to 0,5 wt-%

• The centrifugal pump is operating at a constant flow rate

• The temperature is measured after the pump and is regulated to 20 degrees by the flow rate in the heat exchanger

• The operating speed of the homogenizer, i.e. the flow rate is set from a electrical control panel

• 1:st and 2:nd stage pressures are varied by adjusting the hydraulic valves on the homogenizer

Figure 22: The experimental setup containing a tank 1, centrifugal pump 2, homogenizer 3, heat exchanger 4 and the blue lines represent the piping. The temperature is measured after the pump.

Particles used in experiments

To simulate the suspensions that are processed in the homogenizer particles is added to municipal water. Two different particles are chosen in order to simulate different products with different wear properties.

• Calcium orthophosphate, diameter 10 μm, is used to simulate calcium-enriched milk.

• Corundum, diameter 3 and 10μm, is used to simulate a more abrasive product like chocolate flavored milk.

The particles are circulated for 3-6 hours and then switched for a new batch. This is done to eliminate the influence of the lifetime for the particles,

T

(31)

Operating conditions for homogenizer - added particles

The operating conditions for the experiments are given in Table 6 and Table 7.

Table 6: The seats and forcers used in the experiments for varying gap height with calcium particles. The total homogenization pressure is 280bar and Th=0,2.

Identification number

Q [l/h] h [µm]

S: J1/F: J2 1250 30

S:K2/F:K2 1475 34

S:L1/F:L1 1040 27

S:L2/F:L2 1940 42

S:N2/F:N2 2600 54

S:O2/F:O2 1933 42

Table 7: The forcers and seat used for the other variables and fluid. The homogenization pressure is 280 bar. All test 2 hours except for corundum, which was run 30 min and 3 min for 3 and 10 µm

respectively.

Variable h [µm] Th-number

S:M2/F:M2 Tomato paste 28 0,2

S:Test2/F:Test2 Corundum 3 µm 42 0,2

S:N1/F:N1 Corundum 10 µm 42 0,2

S:O1/F:O1 Cavitation and calcium particles

54 0

S:P1/F:P1 Calcium particle wear

54 0,6

6.3.2 Experimental setup – Municipal water without added particles The cavitation experiments are carried out on the machine types Tetra Alex 200, 350 and 30. The homogenization pressures for the machines are 200, 400 and 630 bar respectively. Municipal water is circulated in the system so the difference compared to the experimental setup at PDC is that there are no added particles. The variable in these experiments is the 2:nd stage pressure, which is changed to vary the Th-number. The procedure for determining the pressure ratio is done in two steps. Step one is to raise the 2:nd stage pressure to a desired value. Step two is to raise the 1:st stage pressure to the total homogenization pressure.

For dimensions see the drawings of the gap for each machine in Table 5.

(32)

Operating conditions for homogenizer - without added particles

Table 8: The flow rate for TA200 and TA30 is 5500l/h for TA350 the flow rate is 13500l/h.

Machine model Th-number

S/F: S1 TA200 0

S/F: S2 TA200 0,1

S/F: R1 TA350 0

S/F: T1 TA350 0,2

S/F: U1 TA30 0

S/F: U2 TA30 0,1

S/F: V2 TA30 0,2

6.4 Particle size distribution

To measure the particle size distribution in a fluid the instrument Beckman coulter LS230 is used. The purpose of these measurements is to see how the wearing particles are affected by the homogenization tests carried out. There are several measuring options available for this measurement the optical method was Fraunhofer and the PIDS value was between 45-52 %. As can be seen in Figure 23 the particle size was almost unchanged and thus it is possible to re-circulate particles for at least 6 hrs.

Figure 23: Green unhomogenized sample, red homogenized for 4 hrs, pink homogenized for 6 hrs.

(33)

6.5 Analysis of the worn surfaces with optical 3-D profiler

The characteristic measures for the worn surface is measured with the Keyence optical profiler, VK-9700 (USA). These measurements give a high quality picture of the worn surface with high accuracy. The data is plotted against the running conditions for the machine.

When the worn surfaces are analyzed using the microscope there are some limitations in the equipment that must be considered. The requirements for scanning surfaces are determined depending on the information needed. The requirements for different measurements are listed in Table 9.

Table 9: Requirements objective for different measurements.

Type of measurement Demanded

magnification Surface height, 3D-picture & 2-D SEM 20 X

Surface roughness measurement 50 X

Two areas on the forcer and seat are determined for analysis before the measurement.

One area is on the outlet side for homogenization device and the other area is on the opposite side, see Figure 24.

The wear on the homogenization device outlet side is measured with 20X magnification and the opposite side with 10X magnification. The idea is to see if the wear pattern is even distributed over the surface, i.e. how frequent scratches and grooves occur.

When measurements on larger areas than the field of view of the objective is carried out the pictures are merged together by the feature “assemble pictures” in the program. The assemble pictures feature is an automated procedure built in to the program.

Engraving

180° from the device outlet Outlet for

the device

Outlet for the device Engraving

Figure 24: Picture illustrating the positions where the analysis is done with the 3d profiler. On the left is the seat and on the right is the forcer. “Outlet for the device” means that it’s the side of the

homogenization device where the outlet is positioned.

(34)

Summary of the procedure for measurement of a worn surface:

1. Scanning of the surface heights with 20 X magnification at 90 degrees.

2. Overview SEM-picture of the wear at 270 degrees.

3. From the 3-d picture chose a representative position and plot the cross section in 2-D.

4. Determination of the characteristic dimensions of the wear pattern.

Scattered wear pattern – localized wear

In Figure 25 picture a) is the unworn surface and picture b) is the worn surface. Yellow lines indicate where the crossection is viewed. In Figure 27 the two crossections are compared to see where the weight loss has occurred. The same procedure is carried out for Figure 26 which gives the crossection in Figure 28.

Figure 25: When the wear pattern is scattered and most weightloss occurs locally in grooves the depth, length and width of the groove is studied. The yellow lines indicate where the crossection is viewed.

(35)

Figure 26: The crossection for determination of the width of the groove.

Figure 27: The maximum height is measured between the two horizontal lines. The length of the scratch is measured between the two vertical lines.

(36)

Figure 28: The width of the groove is measured between the two vertical yellow lines.

Crater wear pattern

The same procedure is carried out for the crater but other variables are of interest. See

XFigure 30.

Figure 29: Comparison of the crossections at the indicated positions. Gives the wear area, center radius, width of the wear and the maximum depth.

(37)

Figure 30: The worn area is measured and afterwards the position of the 50 % of the worn area-radius is measured.

6.6 Determination of weight loss for the gaps

Two balances are used to obtain the weight loss for the gaps because of weight constraints and the need of accuracy. Before weighing a gap it has to be cleaned thoroughly to get an accurate value for the weight loss.

6.6.1 Cleaning gaps

The samples used in the experiments are cleaned before weighing. This is done to assure that the sample is free from metal flakes, which could affect the accuracy of the weighing. The cleaning is done with ethanol 50%. The method starts with immersing the gaps in the alcohol and then brushing them with a dish-brush. The last part of the cleaning process is to wipe clean with a paper. To determine the accuracy of the

cleaning method several fingerprints is applied to the surface and a small piece of paper is weighed with the result of 2 mg accuracy for the method. The method is displayed in Figure 31.

(38)

Figure 31: Cleaning method starting with a) seat immersed in ethanol, b) brush seat in ethanol, c) rigorous cleaning with brush, d) move to balance with pincer or paper

6.6.2 Weighing gaps

The weight loss for the forcer, forcer disk and seat is obtained with two balances. A Metler Toledo XS205 and Metler Toledo LP 1200S.

The XS205 is used for samples weighing lower than 220g due to the upper limit it can measure with precision 0.1mg. The resulting weighting accuracy of the method is estimated to be ±2,1 mg. This measuring error limit is introduced because there might be some fingerprints and paper fibers left on the sample after the cleaning procedure.

The Metler Toledo LP 1200S has a lower precision of 1 mg but a higher weighting

interval and can therefore be used for weighting of the cavitation-erosion samples from the workshop. The resulting weighing accuracy for the method is thus ±3 mg. A pincer is used for moving the samples to the balance to obtain as low external influence as possible on the weight of the sample. When the weight loss is measured a delay of 15 seconds is used to assure a correct value for the weight.

6.6.3 Determination of weight loss with optical profiler

The optical estimation of the weight loss is carried out by using the worn area pointed out in ^XFigure 32^X to determine the worn of volume by multiplying by the periphery. The periphery is calculated using the diameter to the inlet plus the gap length minus the position of the 50 % area for each sample.

b) c) d)

a)

(39)

Figure 32: Definitions of surface profile measures.

6.7 CFD simulations

The simulation software used is Ansys cfx 11. The geometry is generated in the CAD program NX 6 and is modeled as a thin 2-d slice and as the right hand side of the symmetry axis in Figure 9. The grid is generated in cfx-mesh and contains 350000 nodes for the case with outlet chamber and 200000 without outlet chamber see Figure 49 and Figure 50 respectively. The boundary condition (B.C) at the inlet is static pressure. The outlet B.C is set as the mass flow rate calculated by taking the thickness of the slice into account as (thickness of slice/circumference of device)*total mass flow rate. The B.C for the walls are set to no slip and the slip layers is set to free slip. The turbulence model used is k-ε and the differencing scheme is second order accurate. Particle tracking is turned on and the drag model is chosen to be standard. The particle is set to have a spherical shape and a diameter of 10 µm. An erosion model is used to predict the magnitude of the erosion caused by the impact of the particle, the model used is set to finne. The finne model is one of the simplest models to predict erosion.

(40)

7 Result & discussion

The weight loss is measured as the difference between the weight before and after wear for the seat and forcer. Data and shape of surface profiles can be seen in appendix.

7.1 Time evolution of wear

Weight loss

The time evolution of wear is investigated in two experiments. One using calcium particles added to municipal water and one using tomato paste. The weight loss was determined by using the balance method explained in 6.6.2 Weighing gaps. A diagram for the weight loss evolution of calcium particles without cavitation is presented in Figure 33. The wear increases in a linear trend.

Figure 33: Total homogenization pressure 280bar. Total test time is 3 hours.

(41)

In Figure 34 the time evolution of the wear for tomato paste is seen. Initially, first 3 min, the weight loss is very accelerating and afterwards it follows a linear trend.

Figure 34: The trend for the weight loss evolution shows an accelerating trend initially. Afterwards the trend is linear.

7.2 Cavitation erosion

Test series for cavitation are mainly carried out in the workshop except for one test carried out at PDC to see if any differences could be observed between the two experimental rigs.

PDC

The seat and forcer is worn with 280 bars homogenization pressure and Th=0. And as presented in 5.3 Cavitation this is showed to correspond to a point where the cavitation intensity is at a maximum. After 4 hours this setting has not caused any wear that can be noted by visual inspection or weighting. This is consistent with the theory in 5.8 Material theory, the material used gets work-hardened by cavitation through deformation of the material. This explains why it takes some time before the weight loss occurs, since the hardness value is increasing with time until the material fractures thus causing a measureable weight loss.

Workshop

The tests carried out at the workshop are long duration tests (40-600 hours). For a description of the system see6.3.2 Experimental setup – Circulating water^X.

(42)

Weight loss

The weight loss for the three different machines for Th=0 and Th=0,1 is seen in Figure 35. The diameters and operation times vary for the different machines; therefore the weight loss is divided with the diameter of the seat and the operational hours of the test. The weight loss in these tests are calculated as the sum of the weight loss for seat and forcer, this is motivated since the wear mostly affects the seat. The measurement method is the balance method described in section 6.6.2 Weighing gaps. The weight loss seems to be more dependent upon the flow rate or the gap height than the

homogenization pressure. The wear pattern and position of the wear zone for the 600 bar machine for Th=0,2 is seen in Figure 36 and Figure 37. From the figures it can be concluded that the wear zone change position and size with changed Th-number. From Figure 36 it is apparent that wear is minimized for Th=0,1.

Figure 35: The weight loss scale in the diagram is logarithmic and is normalized with the wear time and the radius. The gap heights are 59 µm (200 bar), 94 µm (400 bar) and 72 µm (600 bar).

Figure 36: Damage on the seat for varying Th-numbers. The position and size of the wear damage on the seat a) Th=0 – high cavitation, b) Th=0,1 – medium cavitation, c) Th=0,2 – low cavitation.

a) b) c)

G a p a re a G a p a re a G a p a re a

(43)

Figure 37: Position of the wear zones in the outlet chamber for varying degree of cavitation, 630 bar test.

Wear characteristics

The worn surface in Figure 38 is run with 630 bar of product pressure, the flow rate is 5550 l/h, the induced amount if cavitation is low (Th=0,2) and the product is municipal water.

The outlet has been completely destroyed while the inlet is intact. The remains of the gap area have some grooves, which is probably the effect of limestone that has released from the piping and followed through the homogenization gap. From Figure 38 one can see that cavitation creates most damage to the outlet of the gap, i.e. in the outlet chamber.

Figure 38: Damage on the seat. The grooves at the inlet is due to scratching from particles from the sedimentation in the piping system. At the outlet the typical wear pattern for cavitation erosion is seen

causing the major weight loss.

(44)

7.3 Calcium particles with varying degrees of cavitation

In these tests the influence of cavitation upon particles is tested experimentally.

Weight loss

For the weight loss measurement two methods were used, the balance and the method explained earlier in 6.6.3 Determination of weight loss with optical profiler. The optical weight loss measurement gives an estimate of the weight loss in the gap region but not in the outlet chamber. From ^XFigure 39 it can be concluded that the weight loss is highly dependent upon the Th-number. The seat seems to be more affected by the wear compared to the forcer when analyzing the weight loss obtained from the balance. As seen in Figure 39 the forcer has a higher weight loss compared to the seat for the optical method. The weight loss measured with the balance shows the opposite, therefore it can be concluded that the wear occurs mostly in the outlet chamber for the seat. For the forcer most of the wear occurs in the gap.

One trend that is apparent is the decrease of the weight loss between Th=0 and Th=0,2.

This clearly shows the influence of the Th-number on the weight loss. For lower Th- values the weight loss seems to be higher, this seems to be consistent with the theory that cavitation can accelerate particles in section 5.7 Particle acceleration caused by cavitation.

Figure 39: The gap height is 54 um, total homogenization pressure 280 bar and the wear time 2 hours.

The solid bars represent the balance method and the dashed bars the optical method.

(45)

Wear characteristics

The worn surfaces are run with 280 bar of product pressure, the flow rate is 2600 l/h, the gap height is 54 μm and the product is municipal water with added calcium particles. The differences between the experiments are the amount of cavitation induced. For Th=0,6 which can be seen in Figure 40 there is no cavitation induced. The wear is a result of only the calcium particles added in the fluid. The major difference compared to the cases with cavitation is that the machining marks are still present and the crater formation is not apparent.

Figure 40: Th=0,6, a) forcer, b) seat. The wear on the inlet is not noticable. The machining marks are still left but on a) the forcer they have changed to a smooter pattern and on b) the seat they are smaller . Scratches can be spotted, they are almost as long as the gap but doesnt look deep. The outlet is intact,

only some small damage can be spotted.

a)

b)

(46)

For Th=0,2 seen in ^TFigure 41, a low amount of cavitation is induced. The amount of wear increases when cavitation is induced and most wear appears to be on the forcer when comparing Figure 41 a) and b). A crater has started to form on the forcer and the outlet is worn. The Seat seems to be intact with regards to some radial grooves.

Figure 41: Th=0,2, a) forcer, b) seat. A small crater formation has started on the inlet for a) the forcer, but not on b) the seat. The machining marks are removed in a), but on b) they have become smaller. For a), the outlet is damaged by short and rough grooves, which cannot be seen on b). Instead b) has longer

and thinner scratches at the outlet side.

a)

b)

(47)

For Th=0 a high amount of cavitation is induced and the wear damage can be viewed in

TFigure 42. When raising the amount of cavitation the amount of wear increases. Two craters have now appeared, one on the forcer and one on the seat. The crater on the seat is positioned closer to the outlet when comparing it to the forcer (see Figure 42 a) and b)). There is no outlet damage on the forcer but instead some radial grooves have appeared after the crater.

Figure 42: Th=0, a) forcer, b) seat. The inlet is worn for a) the forcer and for b) the seat there are some scratches. The machining marks are removed for a), but on b) they have become smaller. Crater formation on a) is positioned close to the inlet and b) in the middle of the gap. The crater surfaces are smooth with regards for some black pits. For a), there are large groves after the crater. Some are short

and wide, and some long and narrow. The outlet for a) and b) is intact.

a)

b)

(48)

7.4 Particles with different properties

The wear depending on particle properties was studied experimentally.

Weight loss

In Figure 43, the influence of the properties of the suspended particle can be seen. The difference between the particles is in their mass, hardness and shape. The data shows that wear is highly dependent on particle properties. The largest difference between the particles is that corundum has much higher hardness and that the mass is 3 times

higher. The dominant effect of the corundum particles is quite evident. The weight loss was measured with a balance as explained in section 6.6.2 Weighing gaps.

Figure 43: The weight loss-scale in the diagram is logarithmic. The gap height is 54 µµµµm, Th=0,2 and the homogenization pressure 280bars.

Mechanisms causing wear in a homogenizer gap

M A S T E R ' S T H E S I S